Oracle Books Committed Work 47 Times Faster Than It Delivers It

Oracle spent Thursday evening reporting the largest quarter in its history, and the number that mattered most was not the one on the top line. Total revenue for the first quarter of fiscal 2027 hit a record $19.3 billion, up 30 percent year over year — about $4.4 billion of incremental revenue against the same quarter a year earlier. Over those same twelve months, the pile of contracted work Oracle has signed but not yet delivered grew from $455 billion to $664 billion. The company is adding committed future business roughly forty-seven times faster than it is converting it into recognized revenue.

That ratio has been the central question hanging over Oracle stock for a year, and Q1 was the first report in which management could credibly argue the gap is starting to close rather than widen. Cloud infrastructure revenue reached $7.4 billion, up 121 percent — an acceleration from 93 percent growth in Q4. Total cloud revenue was $11.6 billion, up 62 percent. Non-GAAP operating income rose 31 percent to $8.2 billion, holding operating margin roughly flat at 42 percent even as gross margin fell on data center ramp costs. Non-GAAP earnings per share came in at $1.92, up 30 percent. Shares closed up 6.77 percent at $163.29 on about 2.4 times average volume.

Chief Financial Officer Hilary Maxson, presenting alongside co-chief executives Clay Magouyrk and Mike Sicilia, reached for a single word. If I had to describe this quarter in one word, I think it would be acceleration, she told analysts, noting that Q1 revenue grew sequentially for the first time — historically, she said, a record Q4 was followed by a lighter Q1.

Sequentially, remaining performance obligations rose a more modest $26 billion, with more than $30 billion in new AI cloud contracts signed in the period. Maxson offered the clearest conversion math Oracle has yet given: the company now expects around half of its RPO to convert into sales over the next 36 months — while cautioning that contracts booked in Q1 will not touch capital expenditure or revenue until fiscal 2028 or beyond.

Capacity arrived faster than anyone expected

The operational detail was the genuine surprise. Magouyrk said Oracle delivered 850 megawatts of AI capacity containing more than 300,000 GPUs to customers since the end of Q4 — almost three times what it delivered in all of the prior quarter, and 73 percent of everything it shipped in the entire previous fiscal year. At the Abilene, Texas campus, Oracle delivered 131,000 GPUs, 1.9 times the Q4 figure; six of eight buildings representing 618 megawatts are now in customer hands, with acceptance time compressed to 24 hours. OpenAI's GPT-6 Astra was trained there.

Two numbers inside that deployment speak directly to the bear case on AI hardware depreciation. Fleet-wide GPU utilization held at 97.9 percent. And all capacity that came up for renewal in the quarter was renewed or resold at a 20 percent premium to the prior contract — on hardware mostly four years old or older. In a world where demand exceeds supply, typically prices don't go down, they do go up, Magouyrk said.

Oracle raised full-year guidance to at least $90 billion in revenue, up 34 percent, and $8.10 in non-GAAP EPS; Q2 guidance calls for revenue growth of 30 to 34 percent, cloud growth of 65 to 71 percent, and EPS of $1.85 to $1.93.

The booked-versus-billed problem, and who pays for it

Strip away the records and Oracle's quarter is a cash flow story with an asterisk. Operating cash flow was a record $23 billion. Capital expenditure was $28 billion. Free cash flow was negative $5 billion. Net cash capex, after customer prepayments, was $18 billion. For the full year Oracle still expects $90 billion to $95 billion in gross capex against no more than $70 billion net. It also completed a $20 billion at-the-market equity issuance in its entirety during the quarter — dilution deployed precisely because the alternative was more debt on a balance sheet investors have already flagged.

Bernstein's Mark Moerdler pressed on exactly that financing architecture, and Magouyrk's correction was the most important sentence on the call. I didn't say, and I don't think myself nor Hilary said, that that doesn't require additional CapEx, he said. We said it doesn't require additional cash from Oracle. The distinction is load-bearing. The capital is still being spent; it is routed through supplier financing, customer prepayments, and bring-your-own-hardware deals in which the customer buys the GPUs and rents Oracle's operational skill. It is a clever way to grow a $664 billion backlog without funding it. It also ties a meaningful slice of future revenue to the solvency of counterparties whose own access to capital is, as Magouyrk conceded, a constraint the industry is adapting around.

Deutsche Bank's Brad Zelnick asked the obvious follow-up: when does free cash flow turn positive? Maxson declined to put a date on it, arguing that each of these projects by nature is a strong free cash flow generating project delivering roughly 100 percent conversion to post-tax EBITDA shortly after ramp. Plausible — and unverified, and the largest unpriced variable in the stock.

Oracle has now shown that a non-hyperscaler can erect gigawatt-scale capacity at speed, keep it nearly fully utilized, and re-price four-year-old accelerators upward. It has not shown that $664 billion of paper converts to cash on the schedule its multiple implies.

What to watch

Three things. Whether Q2, due December 14, sustains sequential growth and shows that 36-month conversion curve biting in earnest. Whether October's Investor Day puts a date on the free cash flow crossover Maxson withheld. And whether the New Mexico air permit and Wisconsin grid negotiations stay on the benign path Magouyrk described — he insisted neither site affects fiscal 2027 guidance, and pointedly noted that any plan relying on 100 percent achievement of every deliverable has a name: a bad plan.